Vidu Q4 Preview

Parameters below were verified against the live model directory on November 5, 2026. Check the model directory for current schema, availability, and prices.

Overview

Vidu Q4 Preview Reference-to-Video is a multimodal video generation model developed by Vidu (Shengshu Technology), designed for high-consistency multi-subject reference generation and flexible first-frame/first-last-frame image-to-video generation. It supports up to 15 reference images and 3 reference audio tracks simultaneously, ensuring precise control over character identity, style, and environmental consistency across generated video sequences. The reference mode supports up to 15 reference images and 3 reference audio tracks simultaneously, enabling precise control over multiple characters, objects, and environments within a single generation. This allows for the creation of complex, narrative-driven video content where consistency across subjects is strictly maintained.

The full model guide, live parameter schema, and playground for every model ID are available in the HiPix model directory.

Model IDs

Model ID	Mode	Notes
vidu/q4-preview	Image-to-Video, Text-to-Video, Audio-reference
vidu/q4-preview-global	Image-to-Video, Audio-reference

Parameters

Requests are submitted to POST /v1/videos (video) or POST /v1/images/generations / POST /v1/images/edits (image). Fields marked metadata.* travel inside the metadata object.

vidu/q4-preview

API field	Type	Required	Values / limit
prompt	string	Yes	—
metadata.images	array	Yes	up to 15; ≤50 MB each; images
metadata.sounds	array	No	up to 3; ≤50 MB each; audio
metadata.resolution	enum	Yes	540p, 720p, 1080p, 2K, 4K; default 720p
metadata.aspectRatio	enum	Yes	16:9, 1:1, 9:16, 3:4, 4:3; default 16:9
seconds	number	Yes	4–15 (integer); default 5
metadata.mode	enum	Yes	reference; default reference
metadata.audio	boolean	No	true / false

vidu/q4-preview-global

API field	Type	Required	Values / limit
prompt	string	Yes	—
metadata.images	array	Yes	up to 15; ≤50 MB each; images
metadata.sounds	array	No	up to 2; ≤50 MB each; audio
metadata.resolution	enum	Yes	540p, 720p, 1080p, 2K, 4K; default 720p
metadata.mode	enum	Yes	reference; default reference
seconds	number	Yes	4–15 (integer); default 10
metadata.aspectRatio	enum	Yes	16:9, 1:1, 9:16, 3:4, 4:3; default 16:9
metadata.audio	boolean	No	true / false

Example

A minimal request for vidu/q4-preview:

curl -X POST https://api.apimart.cc/v1/videos \
  -H "Authorization: Bearer $HIPIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "vidu/q4-preview", "prompt": "A kitten running through a sunlit garden", "seconds": 5, "metadata": {"images": [{"url": "https://example.com/reference.png", "type": "reference_image"}], "resolution": "720p", "aspectRatio": "16:9", "mode": "reference"}}'

Poll the task with GET /v1/videos/{task_id} until status is completed or failed.

Limits, availability, and pricing

• Submit only the input types listed for the exact model ID; unsupported inputs are rejected or removed on HiPix.
• See Submit a Video Task or the model directory for the full request contract, and Pricing & Billing for how charges are settled.
• Prices and availability are shown per model ID in the model directory.